There is a new KNIME forum. You can still browse and read content from our old forum but if you want to create new posts or join ongoing discussions, please visit our new KNIME forum: https://forum.knime.com

InChi conversion into 2D structure

Member for

6 years 5 months Nico1990

Hi,

 

I want to convert my Inchi into 2D structures for commodity. Usually I use the Molecule to Indigo node. I have noticed that in some cases, the depictions of amines connected to aromatic rings are weird.

As example, there is this molecule.

In chEMBL: https://www.ebi.ac.uk/chembldb/compound/inspect/CHEMBL261849

In Pubchem: http://pubchem.ncbi.nlm.nih.gov/summary/summary.cgi?cid=11617559

But in Knime: see attached file.

 

This is always the same Inchi in the 3 cases.

Could you explain the difference? Is there a way to get the usual depiction?

Nico

 

Comments
Tue, 03/19/2013 - 07:28

Member for

7 years 8 months

wdi

 

That is a known problem of the NIST InChI library, and neither a problem of Indigo or KNIME. Any software using the original NIST library will exhibit the same problem.

The decoder sometimes fails competely (InChIs are *not* guaranteed to be reversible), or creates weird tautomers (as in this case).

You could probably use a tautomer normalization step, or compute the InChI key (the key is the same for the decoded form and the normal tautomer) and use that, via the NCI Resolver, to get a better standardized form.

 

 

 

Mon, 04/08/2013 - 11:25

Member for

6 years 5 months

Nico1990

Do you plan to implement an AuxInfo option in your node?

Nico

Fri, 04/19/2013 - 03:47

Member for

7 years

mikhail.rybalkin

Hi Nico!

We've added a support for AuxInfo both in the Molecule to Indigo and Indigo to Molecule nodes. Now you can optionally produce AuxInfo in the Indigo to Molecule node, and also you convert this AuxInfo into a molecules using Molecule to Indigo node.

Thank you for your suggestion and would be glad to hear how it works for you!

Best regards,
Mikhail

 

Tue, 04/23/2013 - 04:02

Member for

6 years 5 months

Nico1990

Hello Mikhail.

That's great news! I've made some tests and can't see any error. The molecule of my first message is now correctly depicted.

I won't generalize for every molecules, my dataset is relatively small, but I like what you've done!

Thank you.

Nico

Mon, 05/06/2013 - 02:26

Member for

6 years 5 months

Nico1990

Regarding the stereo layer (b layer), I think it is not taken into account.

 

InChI=1S/C11H11N5O2/c12-11-15-8(10(18)16-11)6-2-4-14-9(17)7-5(6)1-3-13-7/h1,3,13H,2,4H2,(H,14,17)(H3,12,15,16,18)/b8-6+

AuxInfo=1/1/N:12,5,13,6,11,4,10,3,8,15,1,18,14,7,2,17,9,16/rA:18nCNCCCCNCOCCCCNCONN/rB:s1;s2;d3;s4;s5;s6;s7;d8;s8;s4d10;s11;d12;s10s13;s3;d15;d1s15;s1;/rC:5.7151,-2.9878,0;5.7274,-3.8127,0;4.9467,-4.0793,0;4.7035,-4.8677,0;5.3083,-5.4288,0;5.2466,-6.2515,0;4.565,-6.7163,0;3.7767,-6.4731,0;3.2155,-7.0778,0;3.4753,-5.7051,0;3.8878,-4.9906,0;3.3357,-4.3776,0;2.582,-4.7131,0;2.6683,-5.5336,0;4.4518,-3.4192,0;3.6269,-3.4316,0;4.9267,-2.7446,0;6.3752,-2.493,0;

 

and

 

InChI=1S/C11H11N5O2/c12-11-15-8(10(18)16-11)6-2-4-14-9(17)7-5(6)1-3-13-7/h1,3,13H,2,4H2,(H,14,17)(H3,12,15,16,18)/b8-6-

AuxInfo=1/1/N:15,5,16,4,14,6,18,7,2,12,9,10,17,3,8,11,1,13/rA:18nOCNCCCCNCNNCOCCCNC/rB:d1;s2;s3;s4;s5;d6;s7;d8;s9;s9;s7s11;d12;s6;s14;d15;s16;s2d14s17;/rC:.1371,2.6354,0;.3206,1.831,0;1.125,1.6475,0;1.4829,.9042,0;1.125,.1609,0;.3206,-.0227,0;.1371,-.827,0;-.6215,-1.1513,0;-.5476,-1.9729,0;-1.1689,-2.5157,0;.2567,-2.1565,0;.6799,-1.4483,0;1.5015,-1.3744,0;-.3244,.4917,0;-1.109,.2367,0;-1.5939,.9042,0;-1.109,1.5716,0;-.3244,1.3167,0;

 

return me the same stereoisomer.

 

Nico